HUY NGUYEN

One of the first members of Bank-wide centralization of analytics team with new tech stack for distributed ML, large structured & unstructured data. Launched following projects:

• Digital acquisition: driving partnerships with various data providers (One Mount Group, ManuLife, Masan) for NTB users via hybrid CR models and AI-supported eKYC onboarding, acquired 2 millions new customers for the bank.
• Retail business: various conversion models for Bancassurance (Banca), (unsecured) Lending, Credit Card (CC); recommendation model for Next Best Offer, Channel Optimization, Next Best Action, Churn prediction. Additional revenue from models was estimated 20 million USD in
  2021. Double conversion rate in leads of Banca, CC and Bond products.
• Risk: first bank-wide ML underwriting model deployed on AWS; models for fraud and AML detection, resulting in a 70% reduction in operational costs.
• Bank-wide Feature Store: designing and automating feature engineering pipeline to create more than 1000 features used by internal team, BI team and Risk team. Leading team to transition from monolith infrastructure to modern scalable stacks using Kubeflow, Kubernetes, Spark, and later migration into AWS & Databricks cloud environment.
• AI services: fine tuning and deploying public LLMs that serve internal users: Chatbot for bank regulations & policies, SQL converting tool for bank-wide. The number of employees accessing each month is more than 30%.

• Rowquery: an analysis tool that helps data analysts, data engineer, and data scientist to ingest data from multi-datasource, storing it in a data lake, and querying it through Spark SQL interface.
• Built data pipelines, data extraction, BI solutions for particular enterprise needs.

Key member of the team that develops and implements AI features into the company's products. Involved projects:

• Web intelligent (Social media listening): designed and maintained near real-time crawled data pipelines. Deployed 10 AI services applied to production data, encompassing text mining (such as keyword highlighting and Named Entity Recognition) and image information retrieval (including face recognition and landmark detection). Applied graph algorithms for social network analysis.
• Cyber security: clustered cyber security entities based on graph representation and developed anomalies detection on network data.